34 "Journal of Compaj-ative Neurology and Psychology. 



with the exception that the orientation for each position was re- 

 versed. Thus three trials were given each rat per day, and the 

 same orientation for any one position was repeated every other 

 day. Not more than a total of twelve trials was given to any 

 one rat. These varied conditions were designed to eliminate the 

 possibility of learning to react immediately to a given position. 

 An accurate account of the behavior of each rat was taken, includ- 

 ing the changes in direction of movement, the distance traversed, 

 the turnings inside the alleys, partial returns and the position 

 where the rat seemed to pick up its cue. The conduct was noted 

 by two, and sometimes by three observers. In all, 84 tests were 

 made and the results were tabulated in statistical form. 



No noticeable tendency for the rats to start in the direction 

 in which they have been oriented was observed. They are just 

 as likely to turn around immediately and start off in the opposite 

 direction. Neither do they tend to start either toward the food- 

 box, fV, or back toward the original entrance, O. In other words, 

 the direction of starting is apparently a matter of chance entirely. 

 This fact of itself argues the lack of any immediate orientation. 

 The situation in which they have been placed thus does not 

 influence nor determine their conduct at the beginning of the test. 



The movements in the latter part of the period of exploration 

 are determined to some extent: The rats tend to migrate back 

 toward the starting box, O. In 75 per cent of the trials, the 

 cue was picked up somewhere between the position where they 

 were released and the box 0. The rats often explore on both 

 sides of the position at which they are released, but 85 per cent 

 of the distance traversed in the period of exploration is on the 

 side toward 0. This general fact may be difl&cult of explanation, 

 but that some determining influence is at w^ork is too evident to 

 admit of doubt. The following explanation may be suggested as 

 a possibihty. In learning the maze originally, the rats explore 

 for a distance from and retrace their steps. This performance is 

 repeated on successive trials with more extensive excursions. 

 When the rats become lost or confused during any trial, although 

 the maze is partially learned, they always run back toward O. 

 It seems that the maze is learned in sections, as it were, and in 

 case the rats become lost at any time, they are able to retrace their 

 steps to more familiar surroundings. When the rat is now intro- 

 duced at the position .v, and begins to explore, the situation becomes 



